Difficult problem noting the volume of Tweets per day and the monumental difficulties faced by the Library of Congress in the United States when it set about trying to archive the tweets-

http://www.politico.com/story/2015/07/library-of-congress-twitter-archive-119698.html

For healthcare meetings Symplur provide a service which provides metrics and summaries of Tweet activity according to the specific #-

http://www.symplur.com/

This useful online article contains a couple of options which may be of use-

https://www.brandwatch.com/2015/03/top-10-free-twitter-analysis-tools/

Hi,

     how about Using computer technology text mining. the "R project" include some packages can be used  mining text.such as "twitteR" package.

R is useful for said problem